Every facility buyer meets spectrum twice: once in the vendor's spectral power distribution graph, and once in the argument about which spectrum grows the better flower. The argument is loud because the evidence genuinely conflicts. Peer-reviewed trials have placed white light first, last and tied for flower yield, sometimes within the same season's literature. This page runs the decision the way a facility decision actually runs: band by band, each with what it does, the state of the evidence behind it, and the question that settles it in a purchase conversation. Two boundaries frame everything below. First, no potency promises: the controlled work consistently puts cannabinoids on genetics rather than on the light recipe, and this page treats that as a finding rather than an inconvenience. Second, every claim carries its evidence state, settled or mixed or cultivar-specific, because that column is the one vendor pages leave out.
Spectrum in One Table
The decision in one table. Each row is unpacked in the section below it. One habit is worth clearing first: Kelvin is a colour tag, not a recipe. Two fixtures both rated 3000K can differ widely in red content and photon output, which is why the Part 1 CCT entry treats it as a work-comfort figure rather than a plant metric, and why what you ask for is the absolute spectral power distribution rather than the Kelvin tag (the Part 1 SPD entry).
| Band | What it does in flower | Evidence state | What to do |
|---|---|---|---|
| Blue 400–500 nm | Compact structure, thicker leaves, tighter internodes; slows biomass at high shares | Settled on shape, mixed on yield | Treat as the architecture dial; state the share in the quote |
| Green 500–600 nm | Reaches chloroplasts red and blue cannot; the canopy-penetration band as PPFD climbs | Mechanism settled; no standalone recipe | The reason a white base earns its place, for plants and crew |
| Red 600–700 nm | The photosynthetic work band and the cheapest photons per joule; high shares raise photobleaching risk in flower | Settled on photosynthesis; bleaching risk cultivar-dependent | The work band; keep the spectrum broad enough for headroom |
| Far-red 700–750 nm | Emerson synergy with PAR; shade-avoidance stretch; an end-of-day timing cue | Mechanism settled; yield and chemistry cultivar-specific | A channel plus a schedule, not a spec-sheet adjective |
| UV 315–400 nm | A narrow-dose terpene lever; no replicated cannabinoid gain in modern trials | Terpene effect: one narrow window. Cannabinoids: unsettled to negative | Justify by dose and purpose, with the safety question answered |
What Spectrum Cannot Do: Potency and the Flowering Switch
Start with the levers spectrum does not have. In controlled crosses designed to separate genetics from environment, genetics accounted for 80% of the variation in THC and 83% in CBD, while environment accounted for 1.7% and 6% respectively (Campbell, Dufresne and Sabatinos, Cannabis and Cannabinoid Research, 2020). CBC sat between the two, roughly half genetic. That partition is not a footnote; it is the reason the spectrum trials below keep landing where they do.
Read the spectrum literature against it and a pattern emerges. A three-cultivar trial in medical cannabis found the cannabinoid response to spectrum real but cultivar-specific, and weaker for the major cannabinoids than for their common precursor CBGA (Magagnini and colleagues, 2021). A Cornell greenhouse trial across six light treatments found no statistically significant CBD or THC difference between any of them, with individual plants varying from 8 to 12% CBD independently of the treatment. The UV trials, covered below, are three for three in finding no cannabinoid gain. None of this says spectrum does nothing. It says what spectrum does is land on shape, biomass partitioning and marketability, and the buyer who pays a fixture premium for a potency spectrum is buying a claim no controlled trial has reproduced.
The Flowering Switch Is Darkness, Not Colour
Spectrum guides are usually read in a room where someone believes red starts flower, so the correction belongs here. Cannabis is a short-day crop: it begins flowering when the night gets long and, above all, stays uninterrupted. That is a clock, not a colour. The sensor that reads the night is the phytochrome system the far-red section deals with below, which is why spectrum can move the plant's shape inside the flowering cycle without being what opens it: the red-to-far-red ratio entry shows what the ratio does control.
Which makes the dark period a lighting decision after all, just not a spectral one. A leak the eye cannot see, down to a fraction of a micromole that a standard PAR meter does not resolve, reads to the plant as a night that never got long enough: flowering can stall, the plant can revert toward vegetative growth, and a crop meant to finish female can throw pollen-producing structures, the outcome growers call hermaphroditism. None of that is a spectrum problem a better lamp tunes away. It is dark-period hygiene, and the Part 1 light leak entry carries the threshold and the audit.
Blue Light Is the Architecture Dial
Blue is the least controversial lever in the file and the most misused. The mechanism is settled: blue photons drive compact vegetative structure, thicker leaves and tighter internodes, and trial work ranks blue ahead of the red-to-far-red ratio as the compactness cue in cannabis (Magagnini and colleagues, 2021). Push the blue share up and the room grows squatter; push it down and the stretch wins.
Where the trials disagree is the yield consequence, and they disagree loudly. The same 2021 trial put the highest flower yield at a blue-to-red ratio of 1:1 and its white treatment last. Cornell's greenhouse trial put plain white ahead of every red-blue mixture it ran for flower dry weight. A greenhouse trial across three cultivars found no dry-flower difference between white and a 90%-red mix at all, while the red mix grew more compact plants that partitioned more biomass into inflorescences at a measurable cost in photosynthetic efficiency (Carranza-Ramirez and colleagues, 2024). A 2025 trial in CBD-rich genetics found flower dry weight and CBD highest at blue-to-red 5:5 and 2:8.
The honest summary is that blue is a shape dial with an unsettled price. What survives across all of it: vegetative rooms carry more blue for structure, flowering rooms carry less because a high blue share slows biomass at the intensities flower rooms run, and the specific percentages printed on vendor charts trace to no published trial. Treat a quoted recipe as the starting point for a split trial in your own cultivar, not as a specification.
Why a White Base Earns Its Place
A white base is usually sold as a preference. It is better defended as engineering, for two reasons that survive scrutiny.
First, the green photons inside white light do work red and blue cannot. Red and blue are absorbed within the top few cell layers of a leaf; green passes through and scatters deeper, reaching chloroplasts the others never excite (Terashima and colleagues, 2009). At low intensity green is the least efficient photon in the room. As PPFD climbs, the ranking inverts: the upper chloroplasts fed by red and blue are already saturated and shedding energy as heat, while green keeps finding unlit chloroplasts deeper in the leaf and in the canopy below (Zhen and Bugbee's 2023 review of blue, green and red photosynthetic physiology). The same physics runs at room scale, and it is the reason lower leaves under a dense canopy photosynthesise at all.
Second, people work under it. A higher-CRI white makes pest scouting, deficiency spotting and ripeness calls possible for the crew walking the rows, and the Part 1 CRI entry explains why that is a people metric worth having. A narrow red-blue room hides what a white room shows.
The counterweight belongs in the same paragraph as the claim. The white-is-better yield story is not settled: Cornell found white first for flower dry weight, Magagnini's trial put white last, and the LED manufacturer whose three-room trial made the broadest-spectrum case has since written publicly that its earlier all-white recommendation was broader than the data. Read together, the defensible position is narrower than any vendor's: a white base gives up little, buys canopy penetration and crew visibility, and is not a yield guarantee.
Red, and the Photobleaching Ceiling
Red is the work band. It drives photosynthesis, and because a red photon carries less energy than a blue one, the same joule buys more of them, which is why red-heavy fixtures quote the most flattering efficacy figures. The blue-to-red ratio entry shows how to read that number before crediting hardware for it.
Red is also where the spectrum's one visible failure mode lives. Photobleaching is the loss of chlorophyll in developing flower: under a narrow, red-heavy spectrum driven at high intensity, buds at the top of the canopy bleach toward white, the symptom growers call albino buds. The mechanism is over-delivery, a single band pushed past what the tissue it feeds can dissipate. The response is strongly cultivar-dependent, with some genetics never showing it and others bleaching readily. It is mostly a marketability event rather than a chemistry one: reports generally find little effect on yield, terpene profile or potency, but the buds it hits are the top-canopy ones that carry the highest price, and bleached flower sells as damaged.
The design response is a spectrum with headroom and an intensity plan, in that order:
- The broader the spectrum, the more headroom before any single band is over-delivered. This is the practical case for a white base, stated in reverse.
- A red-heavy spectrum in flower needs intensity discipline. The PPFD and DLI targets, and the CO₂ caveat that comes with the top of the range, are on the PPFD and DLI page rather than repeated here.
- Under-canopy bars sit too close to the flower for hanging distance to be the fix, so their spectrum should be broad by design; the flagship under-canopy guide covers the configuration.
- Distance and dimming remain the first two levers when bleaching appears, which is why the grower's word for the symptom, light burn, sits in the Part 1 room-risk group.
Far-Red Does Two Things, and Only One Is on the Clock
Far-red purchases go wrong when the two things far-red does get priced as one.
The first is photosynthetic. Far-red photons from 701 to 750 nm do almost nothing on their own, but mixed with PAR at up to roughly 30 percent of the photon flux they lift canopy photosynthesis by as much as an equal number of PAR photons, the Emerson effect (Zhen and Bugbee, 2020). Past roughly that share the response flattens rather than climbing on, because the effect saturates once the two photosystems are back in balance, so more far-red is not more photosynthesis. Past about 752 nm the effect is gone. This is the physiology behind a fact the Part 1 far-red entry covers: a far-red channel measures lower PPE while doing work the metric does not count, a design trade rather than a defect.
The second is morphological, and it is the one with a schedule attached. Far-red pushes phytochrome toward its inactive form, a falling red-to-far-red ratio reads as a neighbour overhead, and the plant answers with internode elongation: the stretch every flowering room plans for. Steer the ratio down and you buy height on purpose; steer it up and you buy compactness. The phytochrome entry carries the mechanism.
The timing lever sits between the two. An end-of-day far-red pulse at lights-off effectively moves dusk earlier. In the trial that matters for cannabis, Peterswald and colleagues (Scientific Reports, 2025) ran ten-hour photoperiods against twelve-hour controls across high-THC and high-CBD cultivars: THC concentration rose in both high-THC cultivars under the far-red schedules, one cultivar's cannabinoid yield per plant rose by roughly 70% against the twelve-hour control, and a four-hour end-of-day treatment brought flowering forward by about four and a half days. The response did not persist once flowering was induced, and the shorter schedule ran about 5.5% cheaper in power.
The evidence state is asymmetric and worth stating plainly: the morphology and timing effects replicate, the yield and chemistry effects are cultivar-specific and unreplicated at commercial scale. The procurement conclusion follows. Far-red is a channel plus a schedule, not a fixture adjective: it needs a switchable channel and a controller that can fire a pulse at a precise minute, which is what the Part 2 end-of-day far-red entry means by a scheduling feature.
UV Is a Terpene Lever, Not a Potency Lever
UV carries the oldest and least earned reputation in the spectrum file.
Where the belief comes from: THC absorbs UV, early work reported flower THC rising from 25% to 32% as the daily UV-B dose climbed (Lydon and colleagues, 1987), and high-altitude populations carry more of several cannabinoids. On that story, UV is a potency dial and a fixture with a UV channel sells itself.
What the modern controlled trials find is narrower and, for the potency story, negative:
- A UV-B dose-response trial across two indoor cultivars found no commercially relevant benefit: flower yield and cannabinoid concentration did not rise, one cultivar's THC and CBD fell as the dose climbed, and terpene content fell in both (Rodriguez-Morrison, Llewellyn and Zheng, 2021).
- A commercial-scale trial run with a licensed producer found no effect of UV on yield or flower cannabinoids; the only tissue with a higher THC concentration was the sugar leaves, and the THC in those tissues did not change (Llewellyn and colleagues, 2022). The point from that paper worth keeping: no mechanism has been identified for UV raising THC while leaving the other cannabinoids alone.
- A 2024 trial from Humboldt University ran three UV spectra at five intensities and found none of the treatments altered the cannabinoid profile (Huebner and colleagues, 2024). What one narrow setting moved was terpenes: at a 99:1 UV-A to UV-B ratio and the lowest intensity tested, linalool rose 29%, limonene 25% and myrcene 22%, with yield and cannabinoid concentration unchanged. Every setting carrying more UV-B bought smaller leaves.
Read together, the position on present evidence is precise: UV is a terpene lever at a stated dose, not a potency lever. The window is narrow, the cultivar map is largely uncharted, and that argues for a switchable channel rather than a fixed one, because the setting that moved terpenes in one trial is not a specification until it replicates in your genetics.
Two non-negotiables sit before the channel is bought. Safety first: UV-B is a personnel question before it is a plant question, and belongs behind shielding, interlocks and a schedule. Then measurement: ask which channel state the efficacy figure was taken in, because a fixture running its UV channels measures lower while doing work the metric does not credit, and the Part 1 UV entry shows the question to ask.
Five Questions That Settle a Spectrum Claim
Spectrum claims are where the least verifiable language in lighting lives. Five questions close most of the gap:
| The question | Why it settles it |
|---|---|
| Which channel state was the SPD and the efficacy measured in? | A fixture with far-red and UV channels measures differently with them on. The number without the state is not a number: why the channel state belongs on the test report |
| Is the figure at the wall or at the diode? | The diode is a package rating under ideal conditions; the meter sees the wall: the losses in between |
| Who ran the trial, and was it randomised? | University and independent trials sit differently from a vendor's own rooms. Both are usable when labelled, and unusable when passed off as the other |
| What is the denominator: per square metre, per fixture, or per kilowatt-hour? | Spectrum effects are small; the denominator decides whether a difference is a result or noise: two denominators that split LED and HPS |
| What does the PPFD map at my mounting height and dimensions show? | Delivered photons beat brochure photons, whatever the spectrum: how to read a map |
What to Spec: A White Base with Switchable Channels
The configuration the evidence supports is unglamorous: a white-based full spectrum as the platform, the red content as the working band sized to your target PPFD, and far-red and UV as switchable channels with a schedule you own. Everything the trials above reward sits either in the base or in a channel you can turn off, and nothing depends on an unverifiable recipe. The cost side belongs in the same conversation. Red turns a joule of electricity into more photons than any other band does, so each band a fixture adds is a running cost and not a one-time one: the efficacy guide prices that trade and shows how to check the claim.
That is also how the product line is built, which makes the spec conversation short. The 4×4, 4×6 and 4×8 builds ship a white-based full spectrum configured per order, with dual-channel builds adding independent UV and IR control; the Golden series carries UV and IR channels at 3.0–3.5 µmol/J for rooms that run finishing signals; the ECO and Tri-Fold lines carry red, blue, UV and IR in one fixture for the full cycle. Under OEM/ODM the spectrum itself is tuned to your crop protocol, and the controller is what turns a channel into a schedule rather than a switch.
Tell us the cultivar, the target PPFD and the room, and the spectrum that ships is the one the protocol needs, with the channels switchable and the channel states written on the test report.
FAQ
What is the best spectrum for cannabis flowering?
No single spectrum wins across the published trials: peer-reviewed work has placed white light first, last and tied for flower yield. What the evidence supports is a configuration rather than a recipe: a white-based full spectrum for canopy penetration, red as the working band sized to the target PPFD, and blue set for the plant shape the room needs, with far-red and UV as switchable channels justified by dose and schedule. Any vendor chart that prints one percentage set as the answer is tracing to no published trial.
Does the blue-to-red ratio matter for cannabis?
Yes, but mostly as a shape dial. Blue drives compact vegetative structure, thicker leaves and tighter internodes, and trial work ranks it ahead of the red-to-far-red ratio as the compactness cue in cannabis. The yield consequence of a given ratio is where trials disagree: published work has put blue-to-red 1:1 first, found no difference between white and a 90%-red mix, and put white ahead of every mixture it ran. State the ratio, run a split trial in your own cultivar, and treat printed recipes as starting points rather than specifications.
Is full spectrum better than red-enhanced LED for flowering?
The trials genuinely conflict: one found its white treatment lowest-yielding, others found white first or tied with red-heavy mixes. The defensible case for a white base is not yield, it is physics and crew: green photons reach chloroplasts that red and blue never excite, which matters more as PPFD climbs, and white lets the crew see what a narrow-band room hides. Red-heavy spectra also carry the photobleaching risk in flower that broad spectra mostly avoid.
Does UV light increase THC in cannabis?
The controlled evidence says no. A UV-B dose-response trial found flower yield and cannabinoid concentration did not rise, with THC and CBD falling in one cultivar and terpenes falling in both; a commercial-scale trial found no effect at all; and a 2024 trial across three UV spectra and five intensities found no treatment altered the cannabinoid profile. What one narrow UV-A window did move was terpenes, up 22 to 29% with yield unchanged. UV is a terpene lever at a stated dose, not a potency lever.
What does end-of-day far-red do for cannabis?
An end-of-day far-red pulse at lights-off converts the day's remaining phytochrome back to its inactive form, which the plant reads as an earlier dusk. In a ten-against-twelve-hour photoperiod trial it brought flowering forward by about four and a half days, raised THC concentration in two high-THC cultivars, and lifted one cultivar's cannabinoid yield per plant by roughly 70% against the twelve-hour control. The response did not persist once flowering was induced. It is a scheduling feature: it needs a switchable channel and a controller able to fire a pulse at a precise minute.
Does green light help cannabis grow?
Yes, and more as intensity rises. Green is the least efficient photon at low PPFD and the more useful one at high PPFD, because red and blue are absorbed in the top few cell layers of a leaf while green passes through and scatters deeper, reaching chloroplasts the others never excite. The same holds at canopy scale for lower leaves under a dense canopy. Green is also the practical reason a white base is worth having: the crew can see the crop.
What causes photobleaching and bleached buds in cannabis?
A narrow, red-heavy spectrum driven at high intensity degrades the chlorophyll in developing flower and turns top-canopy buds white. The response is strongly cultivar-dependent, and it is mostly a marketability event: reports generally find little effect on yield or chemistry, but the buds it hits carry the highest price. The design fix is spectrum headroom, an intensity plan, and distance or dimming first when it appears.
Should I change the spectrum between veg and flower?
Directionally yes, modestly in practice. Vegetative rooms carry a higher blue share for compact structure; flowering rooms carry less, because a high blue share slows biomass at the intensity flower rooms run. Far-red and UV are the channels that change on a schedule rather than with the stage, and the switch between stages is a dimming and scheduling decision as much as a spectral one. The PPFD and DLI page carries the intensity targets by stage.
Can a grow light change THC potency?
Not by much, on present evidence. In controlled crosses, genetics accounted for 80% of the variation in THC and 83% in CBD, while environment accounted for 1.7% and 6%. Spectrum trials consistently find cannabinoid responses small and cultivar-specific, and the UV trials are three for three in finding no gain. Buy the spectrum for shape, penetration and marketability, buy the genetics for potency, and buy the efficacy for the electricity bill.
What Comes Next
If the spectrum is settled and the intensity is the open question, the PPFD and DLI guide carries the stage-by-stage targets. If the fixture economics are what is unresolved, the LED versus HPS comparison and the efficacy guide run the numbers. And if the words on the quote are the problem, the 30 terms every buyer meets first is the translation layer.
Choosing a Spectrum for Your Room?
Send us the cultivar, the target PPFD and the room: we will come back with a spectrum configuration, a fixture layout and indicative pricing.